The Cauchy problem for a generalized Camassa-Holm equation (Q5891404)

The authors study the initial value problem for the generalized Camassa-Holm equation \[ (q-\partial _{x}^{2})u_{t}=u^{k}u_{3x}+bu^{k-1}u_{x}u_{xx}-(b+1)u^{k}u_{x},u|_{t=0}=u_{0}(x),k\in \mathbb{N}, \tag{*} \] in Sobolev spaces \(H^{s},s>3/2,\;\)\ and showed its well-posedness. Namely, the following theorem holds: Theorem 1. If \(u_{0}(x)\in H^{s},s>3/2,\) on the circle or the line, then there exists \(T=T_{0}(u_{0})>0\) and a unique solution \(u\in C([0,T\};H^{s})\) of (*) which depends continuously on the initial data \(u_{0}.\) Moreover, the solution size and life-span estimations are: \[ \|u(t)\|_{H^{s}}\leq 2\|u_{0}\|_{H^{s}},0\leq t\leq T_{0}\leq (2kc_{s}\|u_{0}\|_{H^{s}}^{k})^{-1}, \] where \(c_{s}>0\) is a constant depending on \(s\). A proof of Theorem 1 is based on a Galerkin type approximation scheme (see [\textit{M. Taylor}, Proc. Am. Math. Soc. 131, No. 5, 1501--1507 (2003; Zbl 1022.35096); Pseudodifferential operators and nonlinear PDE. Boston, MA etc.: Birkhäuser (1991; Zbl 0746.35062)]) and its reducing to an ordinary differential equation problem \(u_{t}=-J_{\varepsilon }[(J_{\varepsilon }u)^{k}J_{\varepsilon }u_{x}]-F(u),\) where \(J_{\varepsilon }\) is a suitable Fridrichs mollifier. In addition, the authors stated the non-uniform continuity of the solution mapping \(H^{s}\ni u(0)\rightarrow u(t)\in C([0,T\};H^{s})\). Similar results were also obtained before in [\textit{F. Tiğlay}, Trans. Am. Math. Soc. 360, No. 4, 1861--1877 (2008; Zbl 1146.35069)], where geometric considerations owing to V. Arnold were effectively elaborated.